(12) United States Patent
`Hoen
`
`(10) Patent N0.:
`(45) Date of Patent:
`
`US 6,253,001 B1
`Jun. 26, 2001
`
`US00625300lBl
`
`(54) OPTICAL SWITCHES USING DUAL AXIS
`MICROMIRRORS
`
`(75)
`
`Inventor: Storrs T. Hoen, Brisbane, CA (US)
`
`(73) Assignee: Agilent Technologies, Inc., Palo Alto,
`CA (US)
`
`( * ) Notice:
`
`Subject to any disclaimer, the term of this
`patent is extended or adjusted under 35
`U.S.C. 154(b) by 0 days.
`
`(21) Appl. No.: 09/488,344
`
`(22)
`
`Filed:
`
`Jan. 20, 2000
`
`Int. Cl.7
`(51)
`(52) U.S. Cl.
`
`G02B 6/26
`
`. . .. ..
`. . . . . . . . . . .. 385/17; 385/16; 385/18;
`385/19
`(58) Field of Search .................................. 385/16, 17, 18,
`385/19, 20, 25, 31; 359/291, 224, 223,
`292
`
`(56)
`
`References Cited
`U.S. PATENT DOCUMENTS
`
`4,754,185
`5,378,954
`5,534,740
`5,872,880 *
`5,900,133 *
`5,986,381
`
`310/309
`6/1988 Gabriel et al.
`310/309
`1/1995 Higuchi et al.
`. 310/309
`7/1996 Higuchi et al.
`2/1999 Maynard iiii..
`385/19 X
`9/1999 Toililinsoil
`385/18
`
`310/309
`11/1999 Hoen et al.
`OTHER PUBLICATIONS
`
`Niino, Toshiki et al., “Dual Excitation Multiphase Electro-
`static Drive,” IEEE, 1995, pp. 1318-1325.
`Niino, Toshiki et al., “Development of an Electrostatic
`Actuator Exceeding 1ON Propulsive Force,” IEEE, 1992,
`pp. 122-1127.
`
`Niino, Toshiki et al., “l-ligh-Power and High-Elliciency
`Electrostatic Actuator,” IEEE, 1993, pp. 136-241.
`Trimmer, W.S.N., “Design Considerations for a Practical
`Electrostatic Micro-Motor," Sensors and Actuators, 11,
`1987, pp. 189-206.
`
`* cited by examiner
`
`Primary Examirzer—Phai1 T. H. Palmer
`
`(57)
`
`ABSTRACT
`
`In a first embodiment of an optical switch having at least one
`dual axis micromirror, the micromirror is manipulated abou
`two generally perpendicular axes by varying Voltage pat
`terns along two electrostatic arrangements. The two elec
`trostatic arrangements may be formed to independently
`drive two movers, or may be formed to control a mover tha
`is displaceable in two directions. The micromirrors and the
`movers that control the micromirrors maybe integrated onto
`a single substrate. Alternatively, the micromirrors may be
`formed on a substrate that is attached to the substrate tha
`includes the mover or movers. In a second embodiment 0
`an optical switch in accordance with the invention,
`the
`switch includes two collimator arrays and two dual axis
`micromirror arrays. Each first micromirror in the first micro
`mirror array is dedicated to one of the collimators in the firs
`collimator array. Similarly, each second micromirror of the
`second niicronnrror array is dedicated to one of the colli
`mators of the second collimator array. By manipulating a
`first micromirror, an input signal from the associated firs
`collimator can be reflected to any of the second micromir-
`rors. By manipulating the second micromirror that receives
`the signal, the signal can be precisely positioned on the
`second collimator that is associated with the second micro-
`mirror.
`
`19 Claims, 12 Drawing Sheets
`
`
`
`FNC 1018
`
`

`
`U.S. Patent
`
`Jun. 26,2001
`
`Sheet 1 of 12
`
`US 6,253,001 B1
`
`

`
`U.S. Patent
`
`Jun. 26,2001
`
`Sheet 2 of 12
`
`US 6,253,001 B1
`
`FIG.3
`
`FIG.2
`
`

`
`U.S. Patent
`
`Jun. 26,2001
`
`Sheet 3 of 12
`
`US 6,253,001 B1
`
`FIG.5
`
`('\l
`
`V‘
`
`I
`
`X
`
`46
`
`40
`
`

`
`1002(092
`
`PI.04tCChS
`
`1B1
`
`o
`
`8
`
` %XWAVOAWOW.O
`
`000
`m0mW..%owmo.M
`
`LLL.
`
`O0
`
`mm
`
`8
`
`'
`
`O0
`
`m.0_n_JN8
`
`x
`
`
`
`

`
`U.S. Patent
`
`Jun. 26,2001
`
`Sheet 5 of 12
`
`US 6,253,001 B1
`
`FIG.7
`
`

`
`U.S. Patent
`
`Jun. 26, 2001
`
`Sheet 6 of 12
`
`US 6,253,001 B1
`
`<1:
`I\
`
`9 L
`
`L
`
`

`
`U.S. Patent
`
`Jun. 26,2001
`
`Sheet 7 of 12
`
`US 6,253,001 B1
`
`76 \‘
`
`86
`
`ZI——>Y
`
`86
`
`FIG. 8
`
`

`
`U.S. Patent
`
`Jun. 26,2001
`
`Sheet 8 of 12
`
`US 6,253,001 B1
`
`>
`35
`
`>
`>>
`>
`Bfigiflfigfi
`
`FIG.9
`
`

`
`U.S. Patent
`
`Jun. 26,2001
`
`Sheet 9 of 12
`
`US 6,253,001 B1
`
`FORCE, ;2N
`
`LATERAL FORCE
`
`OUT-OF-PLANE FORC
`
`MOTOR GAP, /2m
`
`FIG. 11
`
`OUT-OF-PLANE FORCE
`
`LATERAL FORCE
`
`MOTOR GAP, ,um
`
`FIG. 12
`
`

`
`U.S. Patent
`
`Jun. 26,2001
`
`Sheet 10 of 12
`
`US 6,253,001 B1
`
`FIG. 13
`
`

`
`U.S. Patent
`
`US 6,253,001 B1
`
`M2.O_..._
`
`m:Kn#._.%fl_502%EaE
`
`.3.®_..._
`
`©—..U_n_
`
`

`
`U.S. Patent
`
`Jun. 26,2001
`
`Sheet 12 of 12
`
`US 6,253,001 B1
`
`FORM AN ARRAY OF SURFACE ELECTROSTATIC MOVERS
`AND SURFACE ELECTROSTATIC ARRANGEMENTS
`
`FORM AN ARRAY OF DUAL AXIS MICROMIRRORS
`
`SUPPORT THE MICROMIRRORS
`FOR MANIPULATION BY THE MOVERS
`
`TILE MICROMIRROR ARRAYS
`
`POSITION MICROMIRROR AND COLLIMATOR ARRAYS
`
`FIG. 17
`
`134
`
`136
`
`I33
`
`140
`
`142
`
`

`
`US 6,253,001 B1
`
`1
`OPTICAL SVVITCHES USING DUAL AXIS
`MICROMIRRORS
`
`TECHNICAL FIELD
`
`The invention relates generally to optical switches and
`more particularly to optical cross-connected switches having
`micromirrors that are individually manipulated.
`BACKGROUND ART
`
`Continuing innovations in the field of fiberoptic technol-
`ogy have contributed to the increasing number of applica-
`tions of optical fibers in various technologies. With the
`increased utilization of optical fibers, there is a need for
`eflicient optical devices that assist in the transmission and
`the switching of optical signals. At present, there is a need
`for optical switches that direct light signals from an input
`optical fiber to any one of several output optical fibers,
`without converting the optical signal to an electrical signal.
`The coupling of optical fibers by a switch may be
`executed using various methods. One method of interest
`involves employing a micromirror that
`is placed in the
`optical path of an input fiber to refiect optical signals from
`the input fiber to one of alternative output fibers. The input
`and output fibers can be either uni-directional or bidirec-
`tional fibers. In the simplest implementation of the mirror
`method, the input fiber is aligned with one of two output
`optical fibers, such that when the mirror is not placed in the
`optical path between the two fibers, the aligned fibers are in
`a communicating state. However, when the mirror is placed
`between the two aligned fibers,
`the mirror steers (i.e.,
`reflects) optical signals from the input fiber to a second
`output fiber. The positioning of the mirror relative to the path
`of the input fiber can be accomplished by using an apparatus
`that mechanically moves the mirror. There are number of
`proposals to using micromachining technology to make
`optical signals.
`In general,
`the proposals [all
`into two
`categories: in—plane free-space switches and in—plane guided
`wave switches. Free-space optical switches are limited by
`the expansion of optical beams as they propagate through
`free space. For planar approaches, the optical path length
`scales linearly with the number of input fibers. Switches
`larger than 30x30 require large mirrors and beam diameters
`on the order of 1 millimeter (mm). For these planar
`approaches, the number
`of input fibers scales linearly
`with the beam waist and the size of the optical components.
`Thus, the overall switch size grows as N2. It is estimated that
`a 100x100 switch would require an area of 1 ml, which
`would be a very large switch. Moreover, constraints such as
`optical alignment, mirror size, and actuator cost are likely to
`limit the switch to much smaller sizes. One planar approach
`claims that the optical switch can be designed so that it
`scales with the optical path difference, rather than the overall
`optical path. If this is possible, it would certainly allow
`larger switches. IIowever, the optical path difference also
`scales linearly with the number of input fibers for a planar
`approach, so the switch grows very large as it is scaled to
`large fiber counts.
`For guided wave approaches, beam expansion is not a
`problem. Ilowever, loss at each cross point and the difficulty
`of fabricating large guided wave devices are likely to limit
`the number of input fibers in such switches.
`For both approaches, constraints such as loss, optical
`component size, and cost tend to increase with the number
`of fibers. There is a need for an optical cross connect switch
`which scales better with the number of input and output
`fibers. Some free-space optical systems can achieve better
`
`3
`
`10
`
`20
`
`30
`
`mu:
`
`40
`
`S0
`
`60
`
`2
`scaling. These systems make use of the fact that it is possible
`to use optical steering around in two directions to increase
`the optical fiber count. Recently, optical switches that use
`such mirrors have been announced. The systems use piezo-
`electric elements or magnetically or electrostatically actu-
`ated micromirrors. The actuation method for
`these
`approaches is often imprecise. To achieve a variable switch,
`it is typically necessary to use a very high level of optical
`feedback.
`
`What is needed is a micromachine that enables steering of
`optical signals from at least one input to a number of
`alternative outputs, where the arrangement of the outputs is
`not limited to a linear configuration. What is further needed
`is a method of fabricating and arranging arrays of the
`micromachines such that
`the switching is accurate and
`repeatable.
`
`SUMMARY OF THE INVENTION
`
`In one embodiment of an optical switch, a micromachine
`for steering optical signals includes utilizing electrostatic
`forces to manipulate a dual—axis micromirror. The micro-
`mirror is supported adjacent to a substrate to enable move-
`ment of the micromirror relative to the substrate. A first
`surface electrostatic arrangement is configured to generate
`electrostatic forces for rotating the micromirror about a first
`axis. Similarly, a second surface electrostatic arrangement is
`configured to generate electrostatic forces for rotating the
`micromirror about a second axis. The two electrostatic
`arrangements may be used to drive a single mover that
`controls the positioning of the micromirror, or may be used
`to drive separate movers.
`Preferably, an array of micromirrors is formed on a
`substrate. In one application, the micromirrors are formed
`separately from the electrostatically driven movers. For
`example, a micromirror substrate may be formed to include
`an array of micromirrors in a side—by—side relationship, with
`the micromirrors being supported to allow rotation about
`perpendicular first and second axes. The micromirror sub-
`strate may then be attached to a mover substrate on which
`the movers are incorporated, such that the micromirrors are
`generally parallel to the paths of the movers. Each micro-
`mirror may be connected to a projection that extends toward
`the mover substrate and that is controlled by at least one of
`the movers. In this embodiment, the movers manipulate the
`projections in a manner similar to manipulation of joysticks.
`In another embodiment, the micromirrors and movers are
`integrated onto a single substrate. Each micromirror may be
`supported on the substrate by means of a frame. A first
`mover is driven by electrostatic forces to manipulate the
`position of the frame, thereby rotating the micromirror about
`one axis. A second electrostatically driven mover may be
`connected to the micromirror to rotate the micromirror about
`the second axis. However, there may be embodiments in
`which a single mover is used to control rotations about both
`axes. For example, the mover may be electrostatically driven
`in two perpendicular directions.
`Each surface electrostatic arrangement includes at least
`two sets of electrodes. For a particular surface electrostatic
`arrangement, a first set of drive electrodes may be formed
`along a surface of a mover, while a second set of drive
`electrodes is formed along a surface of the substrate. The
`lengths of the electrodes are perpendicular to the direction of
`travel by the mover. The drive electrodes are electrically
`coupled to one or more voltage sources that are used to
`provide an adjustable pattern of voltages to at least one of
`the sets of drive electrodes. The change in the electrostatic
`
`

`
`US 6,253,001 B1
`
`3
`force that results from variations in the voltage patterns
`causes movement of the mover. As an example, the first set
`of drive electrodes may be electrically connected to a
`voltage source that provides a fixed pattern of voltages,
`while the second set is electrically connected to a micro-
`controller that is configured to selectively apply different
`voltages to the individual drive electrodes. The reconfigu-
`ration of the applied voltage pattern modifies the electro-
`static forces between the substrate and the mover, thereby
`laterally displacing the mover.
`Each surface electrostatic arrangement preferably
`includes levitator electrodes on the same surfaces as the
`drive electrodes. Unlike the drive electrodes, the levitator
`electrodes are positioned with the length of the electrodes
`parallel to the direction of travel by the mover. An accept-
`able fixed voltage pattern along the levitator electrodes is
`one that alternates between high and low voltages. Rep11l-
`sive electrostatic forces between the levitator electrodes
`cause the mover to be spaced apart from the substrate. Since
`the levitator electrodes are parallel to the travel direction of
`the mover, the levitator electrodes are not misaligned when
`the mover is displaced laterally. Moreover,
`the repulsive
`electrostatic forces generated between the two sets of levi-
`tator electrodes operate to negate any attractive forces
`generated by the drive electrodes.
`In a separate embodiment of the invention, an optical
`switch is configured to include two separate arrays of dual
`axis micromirrors and two separate arrays of optical signal
`conductors, such as collim ators. One of the arrays of micro-
`mirrors is positioned relative to a first collimator array such
`that each dual axis micromirror is dedicated to one of the
`collimators to receive incident optical signals. The second
`array of micromirrors is positioned relative to the first
`micromirror array to allow an optical signal reflected at the
`first array to be directed to any one of the micromirrors of
`the second array. That is, by manipulating a particular dual
`axis micromirror in the first array, an optical signal incident
`to the particular n1icron1irror can be reflected to any one of
`the micromirrors of the second array. The second collimator
`array is positioned relative to the second array of micromir-
`rors such that the optical signal refleeted by a micromirror of
`the second array is directed to an associated one of the
`collimators in the second collimator array. That
`is,
`the
`micromirrors of the second array are uniquely associated
`with the collimators of the second array, but can be n1anipu—
`lated to provide compensation for the angle of the beam
`from the first array. In this embodiment of the optical switch,
`the manipulation of micromirrors may be accomplished by
`means other than electrostatic forces, without diverging
`from the invention.
`Returning to the embodiment in which the manipulation
`of the micromirrors is implemented by varying generated
`electrostatic forces, a method of fabricating optical micro-
`machines includes forming surface electrostatic movers on a
`surface of a substrate and includes supporting micromirrors
`relative to the substrate such that each micromirror is
`rotatable about substantially perpendicular first and second
`axes and is manipulable by movement of at least one of the
`movers. As previously noted, the movers and the micromir-
`rors may be formed on separate substrates or may be
`integrally fabricated on a single substrate. The movers and
`the mover substrate include the arrays of drive electrodes
`and levitator electrodes. The electrostatic surface actuation
`method is well suited for the positioning of micromirrors
`within the described optical switch, since each micromirror
`may be tilted to approximately 10° on each of the two axes
`and is relatively large from a micromachine perspective. A
`
`4
`micromirror may be on the order of approximately 1 mm
`wide. The mover that drives a micromirror can be displaced
`along actuation distances of approximately 100 ,um, with
`very precise and repeatable positioning. Adequate electro-
`static forces may be generated using voltages of 12 volts or
`lower. The low voltage operation allows the optical switch
`to be coupled with complementary metal-oxide semicon-
`ductor (CMOS) circuitry.
`BRIEF DESCRIPTION OF TIIE DRAWINGS
`
`10
`
`FIG. 1 is a schematic diagram of a 16><16 optical switch
`using dual axis micromirror arrays in accordance with the
`invention.
`
`FIG. 2 is a top view of a schematic representation of a first
`embodiment for positioning two arrays of dual axis 1nicro-
`mirrors in accordance with the invention.
`
`FIG. 3 is a side view of the representation of FIG. 2.
`FIG. 4 is a top view of a second embodiment for posi-
`tioning dual axis arrays of mirrors in accordance with the
`invention.
`
`20
`
`30
`
`inun
`
`40
`
`S0
`
`60
`
`FIG. 5 is a side view of the representation of FIG. 4.
`FIG. 6 is a top view of a micromirror array in accordance
`with one embodiment of the invention.
`FIG. 7 is a side view of one of the micromirrors of FIG.
`6 connected to a mover substrate having actuators for
`manipulating the micromirror about two axes.
`FIG. 7A is a top View that isolates the pair of actuators for
`manipulating the micromirror of FIG. 7.
`FIG. 8 is a bottom view of a mover of FIG. 7, showing
`vertically oriented driver electrodes and horizontally ori-
`ented levitator electrodes.
`FIG. 9 is a side view of the mover and mover substrate of
`FIG. 7, showing voltage patterns along the drive electrodes
`at one particular time.
`FIG. 10 is an end view of one arrangement of levitator
`electrodes on the mover and mover substrate of FIG. 7,
`showing possible voltage patterns along the levitator elec-
`trodes.
`
`in—plane
`FIG. 11 shows graphs of lateral forces (i.e.,
`forces) and out—of—plane forces for surface electrostatic
`drives having a surface area of 1 mm2 and having both drive
`electrodes and levitator electrodes.
`
`in-plane
`FIG. 12 shows graphs of lateral forces (i.e.,
`forces) and out-of-plane forces when the 1 mm: drive
`includes only drive electrodes.
`FIG. 13 is a top view of another embodiment of a
`micromachine having electrostatically driven movers wl1icI1
`manipulate a micromirror about two axes.
`FIG. 14 is a top view of one of the movers and a frame
`of the micromachine of FIG. 13.
`FIG. 15 is a side View of the mover and frame of FIG. 14,
`shown in a rest position.
`FIG. 16 is a side view of the mover and frame of FIG. 15,
`but shown in an operational state.
`FIG. 17 is a process flow of steps for fabricating an optical
`switch in accordance with the invention.
`
`DETAILED DESCRIPTION
`
`With reference to FIG. 1, an optical switch 10 is shown as
`including a first collimator array 12, a second collimator
`array 14, a first micromirror array 16, and a second micro-
`mirror array 18. The optical cross-connect switch utilizes
`dual axis micromirrors to deflect input optical beams to any
`
`

`
`US 6,253,001 B1
`
`5
`one of the output optical elements. In the description of FIG.
`1, the first collimator array 12 will be described as com-
`prising the input elements and the second collimator array
`14 will be described as comprising the output elements.
`However, this is not critical. The individual conductors may
`be bi-directional elements, so that optical signals propagate
`in both directions. Moreover, the use of collimators is not
`critical if other means of controlling beam expansion can be
`substituted.
`A single optical fiber 20 is shown as being connected to
`the first collimator array 12. In practice, there is likely to be
`sixteen optical fibers connected to the 4x4 array. The num-
`ber of elements in the array is not critical to the invention.
`The essential aspect of the optical switch is that each
`micromirror is individually manipulable along two physical
`axes. In FIG. 1, only one micromirror 22 is shown in the llrst
`array 16 and only the two micromirrors 26 and 28 are shown
`in the second array 18. However,
`there is a separately
`manipulable dual axis micromirror for each of the sixteen
`segments of the first array and each of the sixteen segments
`of the second array.
`Each input fiber, such as the fiber 20, is coupled to its own
`collimator in the first collimator array 12. An input optical
`signal 30 from the fiber 20 exits from the collimator array 12
`as a slightly converging beam. The converging beam is
`directed to be incident to a particular micromirror 22 in the
`first niicrotnirror array 16. Thus, each micromirror iii the
`first array is dedicated to one of the collimators. However,
`each micromirror is manipulated to redirect an incident
`beam to any one of the micromirrors in the second array 18.
`For example, the dashed lines from the micromirror 22 of
`the first array 16 to the micromirror 28 of the second array
`18 represents a redirection of the input beam 30 as a result
`of manipulation of the micromirror 22. In the preferred
`embodiment,
`the manipulation of a micromirror, such as
`micromirror 22,
`is achieved using electrostatic forces.
`Nevertheless, other approaches may be employed.
`When the micromirror 22 is pivoted along one of its axes,
`the reflected beam 32 will sweep horizontally across the
`second micromirror array 18. On the other hand, when the
`micromirror 22 is pivoted about its seoond axis, the reflected
`beam 32 will sweep vertically across the second array 18.
`Each of the micromirrors, such as micromirror 26, in the
`second array is dedicated to one of the collimators of the
`second collimator array 14. The dual axis capability of the
`second micromirrors allows each micromirror to be pre-
`cisely positioned, so as to compensate for the angle at which
`the beam arrives from a particular micromirror of the first
`micromirror array 16. Thus, the micromirror 26 is precisely
`positioned about each of its two axes of rotation and
`redirects the optical beam 36 to the corresponding collimator
`34 in the array 14. The rotation of micromirror 26 depends
`upon which micromirror of the first array 16 is directing an
`optical beam to micromirror 26. The optical switch 10 of
`FIG. 1 is symmetrical, so that light beams can pass equally
`eflieiently in either direction.
`As will be explained more fully below, one feature of the
`three-dimensional nature of the design of FIG. 1 is that it is
`possible to easily vary the scale of the optical switch 10 to
`accommodate very large fiber counts. FIG. 2 illustrates a top
`view of an optical switch 38. No particular number of input
`and output ports is intended to be shown in the drawing.
`Rather, FIG. 2 shows the locations of various optical ele-
`ments in order to determine the relationship between the
`width of the collimator array and the maximum optical path
`length. All of the indicated dimensions of the switch are
`referenced to the width
`of the collimator arrays 40 and
`
`10
`
`20
`
`30
`
`inun
`
`40
`
`50
`
`60
`
`6
`42. Also shown in the figure is the longest optical path 44
`that can occur when switching any one of the input colli-
`mators to any one of the output collimators. In this design,
`the longest optical path is 7.3 W. The relationship between
`the longest optical path and the size of the collimator arrays
`places a limit on the number of optical fibers that earl be
`coupled with a particular beam width. Table 1 summarizes
`the constraints placed on the optical switch by the angular
`divergence of a Gaussian beam traveling in free space. The
`parameter \/A characterizes the radial index profile in the
`graded index (GRIN) lens (i.e., n(r)=n,,><(l —Ar2/2)). A suit-
`able manufacturer of graded index lenses is NSG America,
`Inc. in Somerset, N.J.
`
`TABLE 1
`Commercial GRIN Lens Collimators
`
`1.0 mm
`diameter
`./A =
`0.481/'n11n
`99
`
`2.0 mm
`diameter
`/A =
`0.237/nini
`407
`
`4.0 mm
`diameter
`/A =
`0.148/mm
`1041
`
`‘I56
`
`317
`
`507
`
`121 x 121
`
`625 x 625
`
`1156 x 1156
`
`13.2
`
`56
`
`143
`
`6 x 3 X 1.5
`
`25 x 13 X 6
`
`62 x 34 X 15
`
`Paraineter
`Maximum
`symmetrical
`beam length (111111)
`Associated waist
`(Mm)Crossconnect
`size
`Collimater array
`width
`(mm)
`System size
`(1 x w x h, cm3)
`
`For a given collimator, there is a maximum length that an
`optical beam can travel and have the sar11e waist at both ends
`of the beam. This length is called the maximum symmetrical
`beam length in Table 1, and it grows approximately as the
`square of the collimator diameter. Since the optical path in
`the system 38 grows linearly with the collimator diameter, it
`is always possible to achieve larger fiber counts by using
`larger collimators. This [act is borne out in Table 1, where
`1.0 mm diameter collimators can be used to achieve a
`121x121 switch, while 4.0 mm collimators can be used to
`achieve a 1000><1000 switch at the expense of increased
`optical system size. The number of fiber inputs should
`increase approximately as the square of the collimator
`diameter, assuming that the waist of the beam leaving the
`collimator scales as the diameter of the collimator. This is
`not indicated by the three collimators analyzed for Table 1,
`presumably because of the di
`iculties in doping the GRIN
`lenses.
`
`FIG. 3 is a side view of the optical switch 38 of FIG. 2.
`In the two figures, the first and second micromirror arrays 46
`and 48 are shown as being planar devices and individual
`micromirrors are not shown. However,
`the individually
`manipulated tnicromirrors are incorporated into the two
`arrays 46 and 48 so that any one of the input colhinators in
`the collimator array 40 can be optically coupled to any one
`of the collimators in the collimator array 42.
`There are a number of available methods for increasing
`the fiber count for a selected collimator array size. Firstly,
`the system may be made slightly asymmetrical by allowing
`the optical beam to travel more than the maximum sym-
`metrical beam length shown in Table 1. However,
`this
`method has an associated increase in optical losses and
`crosstalk. Secondly, a different switch geometry can be used,
`such as that shown in the top View of FIG. 4 and the side
`
`

`
`US 6,253,001 B1
`
`7
`View of FIG. 5. While the geometry is different, the com-
`ponents are substantially identical, so the reference numerals
`of FIGS. 2 and 3 are also used in FIGS. 4 and 5. In the
`embodiment of FIGS. 4 and 5, the maximum beam length is
`only 4.1 W. 111 this case, the 4.0 111111 GRIN lens could be
`used to create a 3,600x3,600 switch. This syste111 design
`places more difficult requirements on the micromirrors of
`the arrays 46 and 48. Most notably, the micromirrors must
`be able to rotate into the plane of the substrate on which the
`micromirrors are formed.
`
`A third method of increasing the fiber count would be to
`use more efficient collimators for which the output waist is
`a larger fraction of the collimator diameter. Afourth method
`would be to use a c1ose—paeked fiber array, rather than the
`square array shown in FIG. 1. Close packing, however,
`would only increase the nu111ber of optical fibers by 15%,
`and it would make the tiling to be described below 111ore
`difficult to implement. A fifth method is to very accurately
`control the curvature of the micromirrors, so that they can
`operate as focusing elements to compensate for the Gaussian
`beam expansion. A theoretical sixth method would be to use
`optics in the input and output stages, so that the switch
`would scale with the optical path difference, rather than with
`the total optical path length.
`
`REQUIREMENTS ON THE MICRO-OPTICAL
`COMPONENTS
`
`There are a number of constraints which must be
`addressed in the design of an optical switch in accordance
`with the invention. Table 2 summarizes the optical con-
`straints placed on the collimators, micromirrors, and actua-
`tors. Three dillerent size switches are identified in Table 2.
`
`TABLE 2
`
`Commercial GRIN Lens Collimators
`
`1.0 mm
`diameter
`1/A =
`0 .481/1111.11
`
`2.0 mm
`diameter
`/A =
`0. 237/111111
`
`4.0 mm
`diameter
`./A =
`0.148/n1111
`
`2.08
`
`4.22
`
`6.77
`
`121 x 121
`
`625 x 625
`
`1156 x 1156
`
`13.2
`
`21.4
`
`56
`
`20.69
`
`143
`
`20.43
`
`1.0x 0.8
`
`2.1 x 1.6
`
`3.4 x 2.5
`
`1
`
`20
`
`4
`
`20
`
`11
`
`20
`
`20.70
`
`20.34
`
`20.21
`
`21.3
`
`20.89
`
`20.55
`
`20.12
`
`20.06
`
`20.04
`
`Paianieter
`Collimators
`
`Effective focal
`length (mm)
`(frossconnect size
`(input x output)
`Collimator array
`width
`Angular tolerance
`on individual
`colliniators (mrad)
`Micromirrors
`
`Mirror size
`(Lx x LV, mmz)
`Minimum mirror
`radius of curvature
`(111)
`Dynamic angular
`range (degrees)
`Angular precision to
`direct beam to
`mirror on 2”“ array
`(mrad)
`Mirror angular
`precision for
`~40 dB mode
`overlap loss (mrad)
`Mirror angular
`precision for -0.5
`
`8
`
`TABLE 2—conti11ued
`Commercial GRIN Lens Collimators
`
`1.0 111111
`diameter
`IA =
`0.481/mm
`
`2.0 111.111
`diameter
`JA =
`0.237/mm
`
`4.0 111111
`diameter
`IA =
`0.148/mm
`
`150
`
`2300
`
`150
`
`2146
`
`252
`
`226
`
`150
`
`290
`
`217
`
`Parameter
`
`dlj mode overlap
`loss (mrad)
`Actuators
`
`Assumed actuator
`travel (jun)
`Actuator precision
`to direct beam to
`mirror
`on 2*“ array (nrrij
`Actuator precision
`for -0.5 dB mode
`overlap loss (nm)
`
`Regarding the collimators, an ellective focal length equal
`to 1/\/A has been calculated for each GRIN collimator, so
`that it can be compared to standard lenses. Regarding the
`micromirrors, the micromirrors must satisfy Very stringent
`requirements in order to position the optical beams precisely
`on the output collimators, The large beam waists used in the
`switches mean that the mirror sizes must be large, typically
`on the order of several millimeters. Such large mirrors may
`not be possible with so111e of the known surface microma-
`chining techniques used in fabricating micromirrors. With a
`thickness of only a few microns, these known mirrors may
`not be able to maintain the desired flatness (radius of
`curvature) to ensure that the beam propagates without dis-
`tortion. Fortunately, bonded wafer approaches are now
`becoming more common in the manufacture of micro111a-
`chined components, so that it is less di icult to design a
`mirror having a thickness of 100 microns 0 several hundred
`microns. This thiclcness is necessary to ensure that the gold
`film used as a reflective coating on the mirrors does not
`cause undue curvature.
`
`Each micromirror should rotate 10° around two perpen-
`dicular axes in order to couple any input fiber to any output
`fiber. However, the range of 10° may place difficult con-
`straints on other components of the system, such as the
`actuators for 111a11ipulati11g the 111icro111irrors. For the actua-
`tors which will be described fully below, a 10° 111oVe111er1t of
`a 2 mm diameter mirror requires a mover to travel approxi-
`mately 50 to 100 microns. This requirement limits the types
`of micromachined drives that can be utilized. In the pre-
`ferred embodiment, electrostatic surface actuators are 1Iti—
`lized.
`Table 2 also includes three different angular position
`requirements for the micromirrors. An angular precision of
`~05 mrad is required both to position the beam on the
`second micromirror array and to achieve ~40 dB coupling
`(i.e, a maximum overlap loss of ~40 dB) into the output
`fiber. The ~40 dB mode coupling level is selected because a
`sensor could be used to detect this signal level. At this signal
`level, the optical power of the output fiber itself could be
`used to close a control loop which positions the micromir-
`rors. There is a significant benefit in performing the open
`loop control of the beam position on the second micromirror
`array. Otherwise, sensors are required along the area of the
`second micromirror array in order to steer the beam as it
`moves from one micromirror to another. Sensors may also
`be required to ensure that the beam is properly centered on
`the correct output micromirror. Similarly, if the precision for
`
`10
`
`20
`
`30
`
`1»U1
`
`40
`
`S0
`
`60
`
`

`
`US 6,253,001 B1
`
`9
`the ~40 dB mode overlap loss is not met, sensors are
`required to steer the beam onto the correct o11tput collimator.
`Briefly, with regard to the constraints involving the
`actuators, the micromirror properties identified above play
`an important role in determining the requirements of the
`actuators used to drive the micromirrors. For the preferred
`micromirror size and angular range, the actuators must travel
`a distance of approximately 100 microns. An actuator needs
`to be repeatedly positioned with an accuracy of ~0.1 microns
`in order to move the beam between the mirrors on the second
`array, to position the beam in the center of a particular mirror
`in that array, and to achieve ~40 dB coupling into the output
`fiber. This position accuracy can be provided by an electro-
`static surface actuator.
`
`PROPOSED MICROMIRROR DESIGN
`
`FIG. 6 is a top View of an array of sixteen micromirrors
`50 fonned on a micromirror substrate 52. FIG. 7 is a side
`view of one of the micromirrors and the mechanism for
`manipulating the rotations of the micromirror. FIG. 7A is a
`top view of the mechanism for manipulating the micromirror
`rotations. Referring first
`to FIG. 6, each micromirror is
`coupled to a ring member 54 by first and second torsion bars
`56 and 58. The positions of the torsion bars define the first
`axis of rotation of the mirror 50. In the orientation of FIG.
`6, the first axis is a x axis. The ring member 54 is coupled
`to the substrate 52 by third and fourth torsion bars 60 and 62,
`which define the second axis (i.e., the y axis). Only the third
`and fourth torsion bars are visible in the side view of